Diffusion models can be effectively utilized to detect anomalies in telecom network multivariate time-series data, outperforming state-of-the-art baselines.